OF GASES AT HIGH EXHAUSTIONS. 
391 
drawn from a piece of glass rod about 0‘5 millim. in diameter is left attached at 
one end to the rod from which it was drawn, the other end being broken off to the 
right length. This broken end of the fine torsion fibre is inserted into the wider 
end of a fine, somewhat conical, glass tube a, fig. 3 (which must not be much larger 
than is necessary to allow the fibre to enter freely). A small flame applied at b melts 
the glass tube and fibre together. The lower part of the glass tube is bent, as shown in 
the figure, to allow the aluminium support for the mirror, g, and mica plate to be hooked 
on to it. The small piece of glass rod remaining attached to the upper part of the 
torsion fibre is then passed through, so as to project above the top of the glass tube c c, 
fig. 1, which has previously been drawn out to the right size, and held in position by 
an aluminium spring. When the instrument is held in a vertical position, the final 
adjustment can be made by moving the projecting end of the rod; the top aperture is 
then sealed, and the piece of rod carrying the fibre is likewise sealed in at the same 
time. The viscosity apparatus is connected to the pump by the arm h, and tubes i, j, 
k, l, m, n, and o. A flexible glass spiral is introduced at i, so as to allow the apparatus 
to rotate on the pivots b, e, and at the same time to be connected to the pump 
altogether with sealed glass joints. Another arm at p, working between metal stops, 
limits the rotation to the small angle only which is necessary. A spring keeps the arm 
pressed close to one of these stops. 
639. The bulb of the viscosity apparatus is entirely enclosed in a box, packed with 
cotton-wool; in this box is a tube, q, opposite the black half of the mica plate, so that 
a candle at r will shine on it and produce a deflection. This tube is closed at each end 
with glass plates to keep out air-currents, and in front is a shutter to cut off or turn 
on the light of the candle, as desired. The position of the candle flame and its 
distance from the plate are always kept identical in the following manner:—A small 
lens, fixed at some distance from the candle, forms an image of the flame on a white 
screen, and it is so adjusted that when the flame shines direct on to the mica plate 
its image exactly overlaps a mark on the screen. A second system of lens and screen 
is fixed at right angles to the first, and thus the position of the flame can in a few 
moments be adjusted. To avoid unnecessary complication this arrangement is not 
shown in the figure. 
In viscosity experiments when the deflection produced by the candle-flame is not 
measured—as, for instance, in exhaustions below one millimetre—the whole apparatus 
is immersed in a large copper vessel full of water surrounded by a double row of 
Winchester quart bottles, also full of water, and the spaces between them and the 
copper vessel are filled with cotton-wool. A cardboard case surrounds and covers the 
whole. A thermometer reading to tenths of a degree C. is in the water close to the 
bulb. This device, which I have long adopted, and found extremely useful, prevents 
the temperature of the apparatus during a long day’s work from varying more than the 
tenth of a degree. 
The torsional movement given to the mica plate, by the light of the candle shining 
